Lauther/d4-embeddings-v2.0

SentenceTransformer based on intfloat/multilingual-e5-large-instruct

This is a sentence-transformers model finetuned from intfloat/multilingual-e5-large-instruct on the d4-embeddings dataset. It maps sentences & paragraphs to a 1024-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.

Model Details

Model Description

• Model Type: Sentence Transformer
• Base model: intfloat/multilingual-e5-large-instruct
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 1024 dimensions
• Similarity Function: Cosine Similarity
• Training Dataset:
  • d4-embeddings

Model Sources

• Documentation: Sentence Transformers Documentation
• Repository: Sentence Transformers on GitHub
• Hugging Face: Sentence Transformers on Hugging Face

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: XLMRobertaModel 
  (1): Pooling({'word_embedding_dimension': 1024, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
  (2): Normalize()
)

Usage

Direct Usage (Sentence Transformers)

First install the Sentence Transformers library:

pip install -U sentence-transformers

Then you can load this model and run inference.

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("Lauther/d4-embeddings-v2.0")
# Run inference
sentences = [
    'PTE SUZANO',
    'What is a Calibration Record?\nA Calibration Record documents the calibration process of a specific equipment tag, ensuring that its measurements remain accurate and reliable. Calibration is a critical process in maintaining measurement precision and compliance with standards.\n\nKey Aspects of a Calibration Record:\n- Calibration Date: The exact date when the calibration was performed, crucial for tracking maintenance schedules.\n- Certification Number: A unique identifier for the calibration certificate, providing traceability and verification of compliance.\n- Range Values: The minimum and maximum measurement values covered during the calibration process.\n- Calibration Status: Indicates whether the calibration was approved or saved for further review.\n- Associated Units: Specifies the measurement units used in calibration (e.g., °C, psi).\n- Associated Equipment Tag ID: Links the calibration record to a specific equipment tag, ensuring traceability of measurement instruments.\nCalibration records play a fundamental role in quality assurance, helping maintain measurement integrity and regulatory compliance.',
    'What is a flow computer?\nA flow computer is a device used in measurement engineering. It collects analog and digital data from flow meters and other sensors.\n\nKey features of a flow computer:\n- It has a unique name, firmware version, and manufacturer information.\n- It is designed to record and process data such as temperature, pressure, and fluid volume (for gases or oils).',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 1024]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

Training Details

Training Dataset

d4-embeddings

• Dataset: d4-embeddings at 09fb8a5
• Size: 11,165 training samples
• Columns: sentence1, sentence2, and label
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2	label
  type	string	string	int
  details	min: 3 tokens mean: 8.23 tokens max: 19 tokens	min: 27 tokens mean: 187.19 tokens max: 406 tokens	0: ~66.20% 1: ~33.80%

• Samples:

  sentence1	sentence2	label
  Ramal ESVOL - TEVOL (GASVOL 14")	What is Equipment? An Equipment represents a physical device that may be used within a measurement system. Equipment can be active or inactive and is classified by type, such as transmitters, thermometers, or other measurement-related devices. Key Aspects of Equipment: - Serial Number: A unique identifier assigned to each equipment unit for tracking and reference. - Current State: Indicates whether the equipment is currently in use (ACT) or inactive (INA). - Associated Equipment Type: Defines the category of the equipment (e.g., transmitter, thermometer), allowing classification and management. Equipment plays a critical role in measurement systems, ensuring accuracy and reliability in data collection and processing.	0
  Mol (%) CO	What is an Equipment Tag? An Equipment Tag is a unique label string identifier assigned to equipment that is actively installed and in use within a measurement system. It differentiates between equipment in general (which may be in storage or inactive) and equipment that is currently operational in a system. Key Aspects of Equipment Tags: - Equipment-Tag: A distinct label or identifier that uniquely marks the equipment in operation. - Equipment ID: Links the tag to the corresponding equipment unit. - Belonging Measurement System: Specifies which measurement system the tagged equipment is part of. - Equipment Type Name: Classifies the equipment (e.g., transmitter, thermometer), aiding in organization and system integration. The Equipment Tag is essential for tracking and managing operational equipment within a measurement system, ensuring proper identification, monitoring, and maintenance.	0
  FQI-4715-1411	What is a flow computer? A flow computer is a device used in measurement engineering. It collects analog and digital data from flow meters and other sensors. Key features of a flow computer: - It has a unique name, firmware version, and manufacturer information. - It is designed to record and process data such as temperature, pressure, and fluid volume (for gases or oils).	0

• Loss: ContrastiveLoss with these parameters:

  {
      "distance_metric": "SiameseDistanceMetric.COSINE_DISTANCE",
      "margin": 0.5,
      "size_average": true
  }
  

Evaluation Dataset

d4-embeddings

• Dataset: d4-embeddings at 09fb8a5
• Size: 2,392 evaluation samples
• Columns: sentence1, sentence2, and label
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2	label
  type	string	string	int
  details	min: 3 tokens mean: 8.22 tokens max: 19 tokens	min: 27 tokens mean: 183.06 tokens max: 406 tokens	0: ~66.30% 1: ~33.70%

• Samples:

  sentence1	sentence2	label
  PTE UTE JUIZ DE FORA (IGREJINHA) B	What is uncertainty? Uncertainty is a measure of confidence in the precision and reliability of results obtained from equipment or measurement systems. It quantifies the potential error or margin of error in measurements. Types of uncertainty: There are two main types of uncertainty: 1. Uncertainty of magnitudes (variables): - Refers to the uncertainty of specific variables, such as temperature or pressure. - It is calculated after calibrating a device or obtained from the equipment manufacturer's manual. - This uncertainty serves as a starting point for further calculations related to the equipment. 2. Uncertainty of the measurement system: - Refers to the uncertainty calculated for the overall flow measurement. - It depends on the uncertainties of the individual variables (magnitudes) and represents the combined margin of error for the entire system. Key points: - The uncertainties of magnitudes (variables) are the foundation for calculating the uncertainty of ...	1
  measure type	What is a Calibration Record? A Calibration Record documents the calibration process of a specific equipment tag, ensuring that its measurements remain accurate and reliable. Calibration is a critical process in maintaining measurement precision and compliance with standards. Key Aspects of a Calibration Record: - Calibration Date: The exact date when the calibration was performed, crucial for tracking maintenance schedules. - Certification Number: A unique identifier for the calibration certificate, providing traceability and verification of compliance. - Range Values: The minimum and maximum measurement values covered during the calibration process. - Calibration Status: Indicates whether the calibration was approved or saved for further review. - Associated Units: Specifies the measurement units used in calibration (e.g., °C, psi). - Associated Equipment Tag ID: Links the calibration record to a specific equipment tag, ensuring traceability of measurement instruments. Calibration r...	0
  daily flow rate	What is a Measured Magnitude Value? A Measured Magnitude Value represents a DAILY recorded physical measurement of a variable within a monitored fluid. These values are essential for tracking system performance, analyzing trends, and ensuring accurate monitoring of fluid properties. Key Aspects of a Measured Magnitude Value: - Measurement Date: The timestamp indicating when the measurement was recorded. - Measured Value: The daily numeric result of the recorded physical magnitude. - Measurement System Association: Links the measured value to a specific measurement system responsible for capturing the data. - Variable Association: Identifies the specific variable (e.g., temperature, pressure, flow rate) corresponding to the recorded value. Measured magnitude values are crucial for real-time monitoring, historical analysis, and calibration processes within measurement systems. Database advices: This values also are in historics of a flow computer report. Although, to go directl...	1

• Loss: ContrastiveLoss with these parameters:

  {
      "distance_metric": "SiameseDistanceMetric.COSINE_DISTANCE",
      "margin": 0.5,
      "size_average": true
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 12
• per_device_eval_batch_size: 12
• gradient_accumulation_steps: 8
• weight_decay: 0.01
• max_grad_norm: 0.5
• num_train_epochs: 5
• lr_scheduler_type: cosine
• warmup_ratio: 0.1

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: steps
• prediction_loss_only: True
• per_device_train_batch_size: 12
• per_device_eval_batch_size: 12
• per_gpu_train_batch_size: None
• per_gpu_eval_batch_size: None
• gradient_accumulation_steps: 8
• eval_accumulation_steps: None
• torch_empty_cache_steps: None
• learning_rate: 5e-05
• weight_decay: 0.01
• adam_beta1: 0.9
• adam_beta2: 0.999
• adam_epsilon: 1e-08
• max_grad_norm: 0.5
• num_train_epochs: 5
• max_steps: -1
• lr_scheduler_type: cosine
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.1
• warmup_steps: 0
• log_level: passive
• log_level_replica: warning
• log_on_each_node: True
• logging_nan_inf_filter: True
• save_safetensors: True
• save_on_each_node: False
• save_only_model: False
• restore_callback_states_from_checkpoint: False
• no_cuda: False
• use_cpu: False
• use_mps_device: False
• seed: 42
• data_seed: None
• jit_mode_eval: False
• use_ipex: False
• bf16: False
• fp16: False
• fp16_opt_level: O1
• half_precision_backend: auto
• bf16_full_eval: False
• fp16_full_eval: False
• tf32: None
• local_rank: 0
• ddp_backend: None
• tpu_num_cores: None
• tpu_metrics_debug: False
• debug: []
• dataloader_drop_last: False
• dataloader_num_workers: 0
• dataloader_prefetch_factor: None
• past_index: -1
• disable_tqdm: False
• remove_unused_columns: True
• label_names: None
• load_best_model_at_end: False
• ignore_data_skip: False
• fsdp: []
• fsdp_min_num_params: 0
• fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
• fsdp_transformer_layer_cls_to_wrap: None
• accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
• deepspeed: None
• label_smoothing_factor: 0.0
• optim: adamw_torch
• optim_args: None
• adafactor: False
• group_by_length: False
• length_column_name: length
• ddp_find_unused_parameters: None
• ddp_bucket_cap_mb: None
• ddp_broadcast_buffers: False
• dataloader_pin_memory: True
• dataloader_persistent_workers: False
• skip_memory_metrics: True
• use_legacy_prediction_loop: False
• push_to_hub: False
• resume_from_checkpoint: None
• hub_model_id: None
• hub_strategy: every_save
• hub_private_repo: None
• hub_always_push: False
• gradient_checkpointing: False
• gradient_checkpointing_kwargs: None
• include_inputs_for_metrics: False
• include_for_metrics: []
• eval_do_concat_batches: True
• fp16_backend: auto
• push_to_hub_model_id: None
• push_to_hub_organization: None
• mp_parameters:
• auto_find_batch_size: False
• full_determinism: False
• torchdynamo: None
• ray_scope: last
• ddp_timeout: 1800
• torch_compile: False
• torch_compile_backend: None
• torch_compile_mode: None
• dispatch_batches: None
• split_batches: None
• include_tokens_per_second: False
• include_num_input_tokens_seen: False
• neftune_noise_alpha: None
• optim_target_modules: None
• batch_eval_metrics: False
• eval_on_start: False
• use_liger_kernel: False
• eval_use_gather_object: False
• average_tokens_across_devices: False
• prompts: None
• batch_sampler: batch_sampler
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	Validation Loss
0.4296	50	0.1345	-
0.8593	100	0.0512	-
1.2836	150	0.041	0.0051
1.7132	200	0.0344	-
2.1375	250	0.0324	-
2.5671	300	0.0284	0.0038
2.9968	350	0.0296	-
3.4211	400	0.0261	-
3.8507	450	0.0268	0.0035
4.2750	500	0.0244	-
4.7046	550	0.0249	-

Framework Versions

• Python: 3.11.0
• Sentence Transformers: 3.4.1
• Transformers: 4.49.0
• PyTorch: 2.6.0+cu124
• Accelerate: 1.4.0
• Datasets: 3.3.2
• Tokenizers: 0.21.0

Citation

BibTeX

Sentence Transformers

@inproceedings{reimers-2019-sentence-bert,
    title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
    author = "Reimers, Nils and Gurevych, Iryna",
    booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
    month = "11",
    year = "2019",
    publisher = "Association for Computational Linguistics",
    url = "https://arxiv.org/abs/1908.10084",
}

ContrastiveLoss

@inproceedings{hadsell2006dimensionality,
    author={Hadsell, R. and Chopra, S. and LeCun, Y.},
    booktitle={2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'06)},
    title={Dimensionality Reduction by Learning an Invariant Mapping},
    year={2006},
    volume={2},
    number={},
    pages={1735-1742},
    doi={10.1109/CVPR.2006.100}
}